One OpenAI-compatible API key for China's frontier models — a drop-in alternative to official endpoints. Cheaper than official pricing, metered to the token and auditable to the cent. Use Bridge as your DeepSeek API gateway, GLM API gateway, Kimi API gateway or Qwen API gateway — no Chinese phone number, just your email.
Discounted provider channels and self-hosted open weights. DeepSeek V4 Pro at $0.25/$0.65 per 1M tokens — roughly 25% below OpenRouter.
Byte-level OpenAI compatibility: change base_url and api_key, keep every other line. LangChain, Cursor and any OpenAI SDK work unchanged.
Metered by real upstream usage, cached-token discounts applied, every request queryable by its x-request-id. No black-box billing.
What is Bridge? Bridge is an AI API gateway that gives developers one OpenAI-compatible endpoint for China's frontier models — DeepSeek, GLM, Kimi and Qwen — with no Chinese phone number or payment method required.
Is Bridge really OpenAI-compatible? Yes — byte-level compatible. Change base_url and api_key in your OpenAI SDK and keep every other line. Streaming, tool calling and structured outputs work unchanged.
How is billing metered? Per-token, using the upstream usage field as the source of truth, including cached-input discounts. Every request is queryable by its x-request-id.
Which models are available? DeepSeek V4 Pro, DeepSeek V4 Flash, GLM-5.2 (vision), Kimi K3 (256K context), Qwen2.5 72B and Qwen3 embeddings.
How do I get started? Register with just an email address — every new account gets $1 of free credit to try every model.
Models & pricing · API documentation ·{' '} Get started